umbc-scify/PubMedClaim

该数据集是一个用于问题回答的任务，包含三个子集：人工生成的pqa_artificial、带有标签的pqa_labeled和无标签的pqa_unlabeled。每个子集都包括发布ID、问题、上下文信息（包括多个上下文字段、标签和网格）、长答案、最终决策和声明等字段。pqa_artificial子集包含207269个训练样本和2000个验证及测试样本，pqa_labeled和pqa_unlabeled子集各包含500个验证和测试样本，pqa_unlabeled子集额外包含57249个训练样本。

The dataset is a question answering task consisting of three subsets: pqa_artificial which is artificially generated, pqa_labeled with labels, and pqa_unlabeled without labels. Each subset includes fields such as publication ID, question, context information (including multiple context segments, labels, and meshes), long answer, final decision, and claim. The pqa_artificial subset contains 207269 training samples and 2000 validation and test samples each, while the pqa_labeled and pqa_unlabeled subsets each contain 500 validation and test samples, and the pqa_unlabeled subset additionally contains 57249 training samples.